Your comment section is the most honest focus group you will never pay for. Viewers spell out confusion, beg for sequels and name the exact topic they want next.
The catch is volume. Nobody reads 4,000 comments across ten competitor videos and turns them into a clean plan before lunch.
Berpasangan Apify YouTube Comments Scraper with Claude solves exactly that. Apify pulls every public comment into structured data, and the Apify MCP server hands it straight to Claude.
Claude then returns a ready AI video brief with hooks, viewer questions and gaps to fill. No theory here. Just the build, the prompts and the settings that matter.
Quick Mojo Verdict
paling sesuai: Creators, YouTube agencies and faceless channel operators who plan videos from audience demand.
Setup effort: A few minutes inside Claude, zero code.
Kemenangan terbesar: Briefs built on real viewer language, ranked by likes and backed by comment IDs.
Berhati-hati: The scraper pulls just one comment per video by default, and usernames count as personal data.
Bottom line? Few workflows turn raw audience noise into a usable video content strategy this fast.
Inside the Comment to Brief Pipeline (Five Parts, One Job Each)
Every brief passes through the same five stages. Once you see how data flows, each later step makes sense.
Masa latihan
Alat
Output
Find source videos
Apify YouTube Scraper
Video URLs from search terms or channels, with comment counts
Claude launches Actors and reads datasets on its own
Analyze and write
Claude
Themes, questions, objections and a finished brief
Guna semula
Claude Project or Skill
Same brief format on every job
Both scrapers are maintained by Apify itself. The comments Actor sits at a 4.9 rating with roughly 26K users, so you are not betting on an abandoned hobby project.
One more advantage matters a lot. Neither Actor touches YouTube Data API quotas, so a 20 video batch never stalls halfway because Google capped your daily units.
YouTube Comments Scraper Settings That Shape Your Brief
Four input fields control everything. Get them wrong and Claude ends up analyzing a handful of random comments instead of real audience demand.
Bidang
Pilihan
Best setting for briefs
startUrls
Video URLs, or a CSV or Google Sheet import
5 to 15 videos from one topic cluster
maxComments
Limit per video, default is 1
300 to 500 per video
sortCommentsBy
TOP_COMMENTS or NEWEST_FIRST (default)
TOP_COMMENTS for demand signals
oldestCommentDate
Exact date or relative value such as “7 days”
Only for tracking fresh reactions
yang maxComments default of 1 trips up almost everyone. Always state a number in your prompt, or Claude may accept the default silently.
Also note that setting oldestCommentDate auto resets sorting to newest first. Use it for reaction tracking on your own uploads, not for mining evergreen demand.
Output Fields Worth Feeding Claude
Each comment arrives as one clean record. Some fields carry far more strategic weight than others.
Bidang
What it reveals inside a brief
komen
Raw viewer wording for hooks, titles and thumbnail text
voteCount
How many viewers agree, used to rank themes
replyCount
Debates and confusion threads worth a dedicated segment
hasCreatorHeart
Points the creator already validated publicly
authorIsChannelOwner
Creator clarifications, often fixing what the video missed
type and replyToCid
Separates top level opinions from reply chains
publishedTimeText
Relative age, such as “3 days ago”
cid, videoId, title, pageUrl
Source trail so every claim can be checked
Exports come in JSON, CSV, Excel, XML and HTML. With MCP, though, you rarely download anything, since Claude reads the dataset directly.
Connecting Apify MCP Server to Claude (Pick One Route)
Claude needs a bridge to launch Actors. Four routes exist, and most readers only need the first one.
Jalan
Terbaik untuk
Bagaimana ia bersambung
Claude web or Desktop connector
Non technical creators
Search “Apify” in the connector directory, or add https://mcp.apify.com as a custom connector and sign in through OAuth
Claude custom connectors work on Free, Pro, Max, Team and Enterprise plans. Free accounts get one penyambung tersuai, so Apify takes your only slot there.
Team and Enterprise members need an Owner to add Apify first. After that, each member connects individually with their own Apify login.
OAuth means Claude never sees your Apify password or API token. You approve access once in the browser and you are done.
Fetching selected fields keeps Claude's context window lean. Big jobs may outlast one tool call, so Claude receives a dataset ID and checks back automatically.
Step 5. Clean Before You Analyze
Drop items carrying an “error” field, such as NO_COMMENTS or VIDEO_UNAVAILABLE.
Remove spam, emoji only replies and duplicate text.
Replace author handles with neutral labels like Viewer 1.
Step 6. Generate the Brief
Paste the master prompt from the next section. Claude clusters themes, ranks them and writes the brief in one pass.
Before accepting the draft, open the top three questions and check their cited comments. If likes look wrong, ask Claude to recount from voteCount.
Step 7. Save the Format
Store the master prompt inside a Claude Project instruction, or save it as a Skill. Every future brief then follows identical structure without retyping anything.
Master Prompt for Claude Video Brief Generation
Generic prompts produce generic briefs. Give Claude a strict structure and demand evidence for every claim.
You are a YouTube content strategist. Using the comment dataset above,
write a video brief for a new video on [TOPIC] for [CHANNEL NICHE].
Peraturan:
Weight every theme by total voteCount, not by comment frequency alone.
Treat comments with replyCount above 5 as confusion or debate signals.
Cite the cid of 1 to 3 supporting comments after every claim.
Flag sarcasm or jokes you are unsure about instead of guessing.
Brief sections:
1. Core viewer problem in one sentence
2. Top 10 viewer questions, ranked by likes
3. Objections, myths and misconceptions to address
4. Follow up requests (“make a part 2 on…”)
5. What the source videos missed or explained poorly
6. Viewer language bank: 15 exact phrases for titles and hooks
7. Three title options and three opening hook scripts (under 20 words)
8. Section outline with rough runtime per section
9. Pinned comment idea to spark replies after upload
. cid citation rule is what separates a trustworthy brief from confident fiction. Spot check three or four IDs against the dataset before scripting.
Nota operator: brief quality equals question quality. “Summarize these comments” produces mush. “Rank unanswered beginner questions by likes” produces a script.
More Prompts Worth Firing (Swap the Niche)
Same dataset, different jobs. Each prompt below reuses comments you already scraped, so nothing new gets pulled.
From these comments, write 10 YouTube Shorts ideas. Each idea must answer one viewer question in under 45 seconds. Cite the cid.
Compare comments on my video [URL] against comments on [COMPETITOR URL].
Show questions their viewers asked that mine never did.
Pull every comment mentioning a tool, product or price. Group by product and note if viewers sound happy, confused or annoyed.
Turn the 8 most liked questions into an FAQ block I can paste into my video description.
That product mention prompt doubles as affiliate penyelidikan kandungan. Viewers openly name tools they bought, regret or still hunt for.
Reading Comment Signals Like a Content Strategist
Claude does the clustering, but knowing what each signal means helps you push back when a brief feels off.
High likes, low replies: broad agreement. Great raw material for hooks and titles.
High replies, modest likes: confusion or argument. Usually deserves its own video section.
Repeated “part 2” requests: terbukti follow up video ideas with built in demand.
Creator replies correcting viewers: gaps the original script left open.
Hearted comments: apa yang competing creator already considers settled.
Questions repeated across several channels matter most. One viral question on one video is noise; the same question across five creators is real audience pain points data.
Turning Viewer Language into Titles and Hooks
The language bank is where briefs earn their keep. Viewers describe problems in words they also type into YouTube search.
Keep the phrasing raw. “Why does my export look blurry” beats any polished rewrite for search intent.
Open with the most liked frustration. Restating it in the first five seconds signals the video answers it.
Match thumbnail text to repeated words. Two or three words lifted straight from comments keep the promise instantly familiar.
Answer the top confusion thread early. High reply threads show where viewers dropped off in competitor videos.
Ask Claude to mark which title options reuse exact viewer phrases. Those deserve the first YouTube title testing slot.
Scaling Comment Mining Across Competitor Channels
One brief proves the concept. Turning competitor comment analysis into a weekly habit is where content calendars start filling themselves.
Pair Comments with Transcripts for Gap Analysis
Apify YouTube Scraper can also download existing subtitles, or create transcripts through a speech to text option. Feed Claude both transcript and comments together.
Now Claude can compare what the video said against what viewers still asked. That is analisis jurang kandungan built on evidence, not assumptions.
Filter Sources Before Scraping
Search by keyword, then cap regular videos, Shorts and streams separately.
Add upload date filters, from the last hour up to the last year.
Paste channel URLs to scan a competitor's recent uploads.
Automate Fresh Pulls
Apify Schedules relaunch any Actor on a cron cadence, and webhooks can push finished datasets into Slack, Make, Zapier atau Google Drive.
Prefer staying inside Claude? Add the schedules tool category to your MCP URL and Claude can create schedules for you conversationally.
For your own uploads, a weekly pull with oldestCommentDate set to “7 days” surfaces fresh questions while the video still gets traffic.
Answer them in a pinned reply or a quick follow up Short.
Send every weekly result to one Google Sheet. Over time it becomes a viewer question bank that Claude can mine before any new brief.
Quick Sentiment Without a Full Brief
Apify's Comments Analyzer Agent scores YouTube, TikTok and Instagram comments as positive, negative or neutral. It also red flags toxic replies, which helps with launch week monitoring.
Bulk Briefs with Claude Code
Agencies handling multiple clients should install the Apify plugin in Claude Code. It bundles the pelayan MCP and Apify Agent Skills, including apify-ultimate-scraper.
Prefer wiring only the comments Actor? One terminal command does it.
claude mcp add –transport http apify “https://mcp.apify.com/?tools=actors,docs,streamers/youtube-comments-scraper”
From there, one instruction can scrape 50 videos and save a separate brief file per topic cluster.
The Honest Part (Accuracy, Privacy and Limits)
Comment data is messy and personal. Five truths before you scale this past a single test brief.
1. Usernames are personal data. Strip handles before analysis and never republish them in scripts or thumbnails.
2. Top comments skew early. TOP_COMMENTS favors viewers who arrived in the first days. Mix in NEWEST_FIRST when tracking shifting opinion.
3. Dates stay relative. publishedTimeText says “2 weeks ago,” not a fixed date. Log your scrape day yourself before comparing batches.
4. Bots and empty sections distort results.Identical text across accounts inflates themes, and videos with disabled comments return error items instead of data.
5. Claude can misread sarcasm. The cid rule plus a quick manual check catches most errors before they reach a script.
Apify's scrapers collect only publicly shared data, never emails or locations. GDPR still applies once personal information enters your workflow, so collect with a legitimate purpose.
Who Gets the Most Out of This Workflow
Padanan yang kuat
Faceless channels planning scripts from proven demand
YouTube agencies producing briefs for several clients
Course creators hunting for recurring beginner confusion
SaaS marketing teams mining tutorial feedback for product content ideas
Cari di tempat lain
Brand new niches where videos carry under 50 comments
Anyone expecting fully automatic scripting and publishing
Channels needing sentiment scores only, where Comments Analyzer Agent alone does the job
FAQs About YouTube Comments to AI Video Briefs
Do I need coding skills to build this workflow?
The Claude connector route needs only an Apify account, OAuth approval and plain English prompts.
Can Claude read YouTube comments without Apify?
Not reliably. Comment threads load through scripts after the page opens, so plain page fetching usually misses most of them.
Does the scraper capture replies too?
Each record carries a type field marking comment or reply, plus replyToCid linking replies to parents.
Will YouTube API limits block large batches?
Apify's comment scraper works outside YouTube Data API quotas, so batch size depends on your settings instead.
How many comments make a solid brief?
Roughly 300 to 500 top comments per video across 5 to 15 videos gives Claude enough pattern density.
Bottom Line on Turning YouTube Comments into Videos Briefs
Most video ideas start as guesses dressed up as strategy. Comment sections replace guessing with thousands of viewers explaining, in their own words, what confused them and what they want next.
Apify handles extraction at scale, and the MCP server removes every copy and paste step. Claude then turns raw chatter into a structured brief backed by comment IDs.
Set maxComments properly, strip usernames and demand citations. Do that, and your next upload starts with proof instead of a hunch.
Your viewers already wrote the next brief. Read it properly, then build.
Tukarkan Mana-mana Foto kepada Persembahan Gaya Bebas Bulin 47 yang Viral dalam Beberapa Saat
. AI penjana video yang dibina untuk satu pemandangan jalanan yang ikonik.
Tingkatkan Jangkauan E-mel Sejuk Anda dan Tukarkan Bakal Pelanggan Menjadi Pendapatan Secara Autopilot
Platform penglibatan jualan semua-dalam-satu yang dibina untuk prospek B2B volum tinggi
. AI Alat Yang Menukar Satu Foto Produk Menjadi Penyenaraian Amazon Penuh Dalam Beberapa Minit
Penjana penyenaraian, imej dan video Amazon anda semua dalam satu
AI Ruang Kerja Akademik Yang Menukarkan Ringkasan Tugasan Menjadi Draf Sedia Untuk Diserahkan
AI penulisan tugasan, pengesanan dan pemanusiaan yang dikuasakan
Tukarkan Sebarang Ringkasan Reka Bentuk kepada HTML Sedia Pengeluaran dengan AI Ejen
AI ruang kerja reka bentuk yang menggantikan saluran kreatif anda daripada penyelidikan kepada kod yang dihantar
Pengikis Komen YouTube + Claude: AI Aliran Kerja Ringkas Video
Your comment section is the most honest focus group you will never pay for. Viewers spell out confusion, beg for sequels and name the exact topic they want next.
The catch is volume. Nobody reads 4,000 comments across ten competitor videos and turns them into a clean plan before lunch.
Berpasangan Apify YouTube Comments Scraper with Claude solves exactly that. Apify pulls every public comment into structured data, and the Apify MCP server hands it straight to Claude.
Claude then returns a ready AI video brief with hooks, viewer questions and gaps to fill. No theory here. Just the build, the prompts and the settings that matter.
Quick Mojo Verdict
Bottom line? Few workflows turn raw audience noise into a usable video content strategy this fast.
Inside the Comment to Brief Pipeline (Five Parts, One Job Each)
Every brief passes through the same five stages. Once you see how data flows, each later step makes sense.
Both scrapers are maintained by Apify itself. The comments Actor sits at a 4.9 rating with roughly 26K users, so you are not betting on an abandoned hobby project.
One more advantage matters a lot. Neither Actor touches YouTube Data API quotas, so a 20 video batch never stalls halfway because Google capped your daily units.
YouTube Comments Scraper Settings That Shape Your Brief
Four input fields control everything. Get them wrong and Claude ends up analyzing a handful of random comments instead of real audience demand.
yang maxComments default of 1 trips up almost everyone. Always state a number in your prompt, or Claude may accept the default silently.
Also note that setting oldestCommentDate auto resets sorting to newest first. Use it for reaction tracking on your own uploads, not for mining evergreen demand.
Output Fields Worth Feeding Claude
Each comment arrives as one clean record. Some fields carry far more strategic weight than others.
Exports come in JSON, CSV, Excel, XML and HTML. With MCP, though, you rarely download anything, since Claude reads the dataset directly.
Connecting Apify MCP Server to Claude (Pick One Route)
Claude needs a bridge to launch Actors. Four routes exist, and most readers only need the first one.
Claude custom connectors work on Free, Pro, Max, Team and Enterprise plans. Free accounts get one penyambung tersuai, so Apify takes your only slot there.
Team and Enterprise members need an Owner to add Apify first. After that, each member connects individually with their own Apify login.
OAuth means Claude never sees your Apify password or API token. You approve access once in the browser and you are done.
Step by Step Guide to Your First AI Video Brief
Here is the exact sequence. Total hands on time is short; most of the work is Claude doing the reading.
Step 1. Connect Apify Inside Claude
Step 2. Switch the Connector On for Your Chat
Click the “+” button in the chat box, open Connectors and toggle Apify on. Ask Claude “What Apify tools do you have?” to confirm everything loaded.
Step 3. Choose Videos Worth Mining
Pick 5 to 15 videos covering one topic, ideally competitor uploads with heavy comment counts. Let Claude find them if you prefer.
Use Apify YouTube Scraper to search “AI editing video for beginners”.
Return the 10 regular videos (no Shorts) with the highest comment counts, showing title, channel, and URL.
Step 4. Launch the Comment Scrape
Use streamers/youtube-comments-scraper on these URLs.
Settings: maxComments 400, sort Comments By TOP_COMMENTS.
Include replies, and set a spending cap with maxTotalChargeUsd.
When finished, fetch only these fields from the dataset:
cid, videoId, title, comment, voteCount, replyCount, hasCreatorHeart, authorIsChannelOwner, type.
Fetching selected fields keeps Claude's context window lean. Big jobs may outlast one tool call, so Claude receives a dataset ID and checks back automatically.
Step 5. Clean Before You Analyze
Step 6. Generate the Brief
Paste the master prompt from the next section. Claude clusters themes, ranks them and writes the brief in one pass.
Before accepting the draft, open the top three questions and check their cited comments. If likes look wrong, ask Claude to recount from voteCount.
Step 7. Save the Format
Store the master prompt inside a Claude Project instruction, or save it as a Skill. Every future brief then follows identical structure without retyping anything.
Master Prompt for Claude Video Brief Generation
Generic prompts produce generic briefs. Give Claude a strict structure and demand evidence for every claim.
You are a YouTube content strategist. Using the comment dataset above,
write a video brief for a new video on [TOPIC] for [CHANNEL NICHE].
Peraturan:
Brief sections:
1. Core viewer problem in one sentence
2. Top 10 viewer questions, ranked by likes
3. Objections, myths and misconceptions to address
4. Follow up requests (“make a part 2 on…”)
5. What the source videos missed or explained poorly
6. Viewer language bank: 15 exact phrases for titles and hooks
7. Three title options and three opening hook scripts (under 20 words)
8. Section outline with rough runtime per section
9. Pinned comment idea to spark replies after upload
. cid citation rule is what separates a trustworthy brief from confident fiction. Spot check three or four IDs against the dataset before scripting.
Nota operator: brief quality equals question quality. “Summarize these comments” produces mush. “Rank unanswered beginner questions by likes” produces a script.
More Prompts Worth Firing (Swap the Niche)
Same dataset, different jobs. Each prompt below reuses comments you already scraped, so nothing new gets pulled.
From these comments, write 10 YouTube Shorts ideas. Each idea must answer one viewer question in under 45 seconds. Cite the cid.
Compare comments on my video [URL] against comments on [COMPETITOR URL].
Show questions their viewers asked that mine never did.
Pull every comment mentioning a tool, product or price. Group by product and note if viewers sound happy, confused or annoyed.
Turn the 8 most liked questions into an FAQ block I can paste into my video description.
That product mention prompt doubles as affiliate penyelidikan kandungan. Viewers openly name tools they bought, regret or still hunt for.
Reading Comment Signals Like a Content Strategist
Claude does the clustering, but knowing what each signal means helps you push back when a brief feels off.
Questions repeated across several channels matter most. One viral question on one video is noise; the same question across five creators is real audience pain points data.
Turning Viewer Language into Titles and Hooks
The language bank is where briefs earn their keep. Viewers describe problems in words they also type into YouTube search.
Ask Claude to mark which title options reuse exact viewer phrases. Those deserve the first YouTube title testing slot.
Scaling Comment Mining Across Competitor Channels
One brief proves the concept. Turning competitor comment analysis into a weekly habit is where content calendars start filling themselves.
Pair Comments with Transcripts for Gap Analysis
Apify YouTube Scraper can also download existing subtitles, or create transcripts through a speech to text option. Feed Claude both transcript and comments together.
Now Claude can compare what the video said against what viewers still asked. That is analisis jurang kandungan built on evidence, not assumptions.
Filter Sources Before Scraping
Automate Fresh Pulls
Apify Schedules relaunch any Actor on a cron cadence, and webhooks can push finished datasets into Slack, Make, Zapier atau Google Drive.
Prefer staying inside Claude? Add the schedules tool category to your MCP URL and Claude can create schedules for you conversationally.
For your own uploads, a weekly pull with oldestCommentDate set to “7 days” surfaces fresh questions while the video still gets traffic.
Answer them in a pinned reply or a quick follow up Short.
Send every weekly result to one Google Sheet. Over time it becomes a viewer question bank that Claude can mine before any new brief.
Quick Sentiment Without a Full Brief
Apify's Comments Analyzer Agent scores YouTube, TikTok and Instagram comments as positive, negative or neutral. It also red flags toxic replies, which helps with launch week monitoring.
Bulk Briefs with Claude Code
Agencies handling multiple clients should install the Apify plugin in Claude Code. It bundles the pelayan MCP and Apify Agent Skills, including apify-ultimate-scraper.
Prefer wiring only the comments Actor? One terminal command does it.
claude mcp add –transport http apify “https://mcp.apify.com/?tools=actors,docs,streamers/youtube-comments-scraper”
From there, one instruction can scrape 50 videos and save a separate brief file per topic cluster.
The Honest Part (Accuracy, Privacy and Limits)
Comment data is messy and personal. Five truths before you scale this past a single test brief.
1. Usernames are personal data. Strip handles before analysis and never republish them in scripts or thumbnails.
2. Top comments skew early. TOP_COMMENTS favors viewers who arrived in the first days. Mix in NEWEST_FIRST when tracking shifting opinion.
3. Dates stay relative. publishedTimeText says “2 weeks ago,” not a fixed date. Log your scrape day yourself before comparing batches.
4. Bots and empty sections distort results. Identical text across accounts inflates themes, and videos with disabled comments return error items instead of data.
5. Claude can misread sarcasm. The cid rule plus a quick manual check catches most errors before they reach a script.
Apify's scrapers collect only publicly shared data, never emails or locations. GDPR still applies once personal information enters your workflow, so collect with a legitimate purpose.
Who Gets the Most Out of This Workflow
FAQs About YouTube Comments to AI Video Briefs
Do I need coding skills to build this workflow?
The Claude connector route needs only an Apify account, OAuth approval and plain English prompts.
Can Claude read YouTube comments without Apify?
Not reliably. Comment threads load through scripts after the page opens, so plain page fetching usually misses most of them.
Does the scraper capture replies too?
Each record carries a type field marking comment or reply, plus replyToCid linking replies to parents.
Will YouTube API limits block large batches?
Apify's comment scraper works outside YouTube Data API quotas, so batch size depends on your settings instead.
How many comments make a solid brief?
Roughly 300 to 500 top comments per video across 5 to 15 videos gives Claude enough pattern density.
Bottom Line on Turning YouTube Comments into Videos Briefs
Most video ideas start as guesses dressed up as strategy. Comment sections replace guessing with thousands of viewers explaining, in their own words, what confused them and what they want next.
Apify handles extraction at scale, and the MCP server removes every copy and paste step. Claude then turns raw chatter into a structured brief backed by comment IDs.
Set maxComments properly, strip usernames and demand citations. Do that, and your next upload starts with proof instead of a hunch.
Your viewers already wrote the next brief. Read it properly, then build.
AiMojo Mengesyorkan:
Menyertai Aimojo Puak!
Sertai 76,000+ ahli untuk mendapatkan petua orang dalam setiap minggu!
BONUS: Dapatkan $200 kamiAI Mastery Toolkit” PERCUMA apabila anda mendaftar!
Tukarkan Mana-mana Foto kepada Persembahan Gaya Bebas Bulin 47 yang Viral dalam Beberapa Saat . AI penjana video yang dibina untuk satu pemandangan jalanan yang ikonik.
Tingkatkan Jangkauan E-mel Sejuk Anda dan Tukarkan Bakal Pelanggan Menjadi Pendapatan Secara Autopilot Platform penglibatan jualan semua-dalam-satu yang dibina untuk prospek B2B volum tinggi
. AI Alat Yang Menukar Satu Foto Produk Menjadi Penyenaraian Amazon Penuh Dalam Beberapa Minit Penjana penyenaraian, imej dan video Amazon anda semua dalam satu
AI Ruang Kerja Akademik Yang Menukarkan Ringkasan Tugasan Menjadi Draf Sedia Untuk Diserahkan AI penulisan tugasan, pengesanan dan pemanusiaan yang dikuasakan
Tukarkan Sebarang Ringkasan Reka Bentuk kepada HTML Sedia Pengeluaran dengan AI Ejen AI ruang kerja reka bentuk yang menggantikan saluran kreatif anda daripada penyelidikan kepada kod yang dihantar